Deepfake detection
Flag AI-generated or manipulated faces in images and video.
Fraud & Deepfake Prevention
Fraud & Deepfake Prevention
AI-generated faces and manipulated video are an increasingly common way to pass identity checks that rely on a single photo or a basic liveness prompt. FacePlugin pairs liveness detection, document anti-spoofing, and purpose-built deepfake detection to catch these attempts before they reach a decision.
Catches AI-generated facesBlocks replay & mask attacksDocument + face coverage
✓
Face liveness detection
Catch printed photos, video replay, and mask-based presentation attacks.
✓
Document liveness detection
Detect tampered, photocopied, or screen-displayed ID documents.
✓
Overview
Defending Against Synthetic Identity, Not Just Photo Spoofing
Older presentation-attack defenses were built mainly to catch printed photos and simple video replays held up to a camera. Deepfake and face-swap tools have raised the bar: a convincing synthetic face can now be generated or injected in real time, during a live video call or a mobile capture, well enough to fool a weak liveness check.
FacePlugin addresses this with two complementary layers: the Face Liveness Detection SDK (passive and active liveness against printed photos, video replay, and masks) and the dedicated Deepfake Detection SDK, which is built specifically to flag AI-generated or manipulated faces rather than only classic presentation attacks. For document-based flows, the ID Document Liveness Detection SDK adds the same kind of defense against tampered or digitally recreated ID documents.
ID Document Liveness Detection SDK adds the same kind of defense against tampered or digitally recreated ID documents.
Built From
The SDKs Behind This Use Case
Licensed once, run on infrastructure you control, with no per-check fees.
Deepfake detection
Flag AI-generated or manipulated faces in images and video.
Face liveness detection
Catch printed photos, video replay, and mask-based presentation attacks.
Document liveness detection
Detect tampered, photocopied, or screen-displayed ID documents.
How It Works
A Typical Flow
1
Capture
A face image or video, and optionally an ID document, is captured during the flow you’re protecting.
2
Analyze
Liveness and deepfake models independently score the capture for signs of spoofing or AI generation.
3
Flag or Pass
Your application receives a structured result to approve, reject, or escalate to manual review.
FAQ
Frequently Asked Questions
Can this catch a deepfake used in a live video call, not just a photo?
The Deepfake Detection SDK is built to analyze both images and video for AI-generation and manipulation signals, which covers injected or synthetic video used during a live capture, not only still photos.
Do we need both the liveness SDK and the deepfake SDK?
They cover different, overlapping threats – liveness detection focuses on physical presentation attacks (photos, replays, masks), while deepfake detection focuses on AI-generated or digitally manipulated faces. Using both gives broader coverage than either alone, though your risk profile may call for just one.
Does this prevent all fraud, or just face-based spoofing?
It specifically defends the biometric and document-capture steps against spoofing and synthetic media. It’s one layer in a broader fraud-prevention program – it doesn’t replace transaction monitoring, device fingerprinting, or behavioral fraud signals your team may also use.
Ready to Add Fraud & Deepfake Prevention?
Talk to our team about your workflow — we’ll point you to the right SDK combination and get a 15-day evaluation license started.